Chapter 7. J.A. van Herwaarden L.W. Bartels B.E. Muhs K.L. Vincken M.Y.A. Lindeboom A. Teutelink F.L. Moll H.J.M. Verhagen

Transcription

1 Dynamic magnetic resonance angiography of the aneurysm neck: Conformational changes during the cardiac cycle with possible consequences for endograft sizing and future design J.A. van Herwaarden L.W. Bartels B.E. Muhs K.L. Vincken M.Y.A. Lindeboom A. Teutelink F.L. Moll H.J.M. Verhagen Journal of Vascular Surgery 2006;44:22-28

2 Abstract Objective: Proper proximal fixation and stent-graft sealing within the aneurysm neck are critical for endovascular aneurysm repair (EVAR) durability. Computed tomography angiography (CTA) is the gold standard for preoperative sizing of endograft diameters, but the accuracy of these measurements is uncertain because they rely on static images of a dynamic process. The aortic configuration and diameter may change during the cardiac cycle. We studied these phenomena using dynamic electrocardiograph-triggered magnetic resonance angiography (MRA). Methods: Eleven consecutive EVAR patients were included. Dynamic MRA was used to perform preoperative and postoperative measurements. Changes were measured in transverse aortic sections 10 mm below the lowest renal artery (level A), at the level of the renal arteries (level B), and 3 cm above the lowest renal artery (level C). Data were analyzed using image segmentation software. Aortic area and diameter changes along 256 axes were determined. Results: Dynamic MRA demonstrated significant aortic area changes during the cardiac cycle before and after EVAR at all three measured levels. Pre-EVAR aortic area significantly increased per cardiac cycle: 8.4% at level A; 9.3% at level B; and 13.3% at level C (P<.001 for all levels). Post-EVAR aortic area increased 9.7% at level A, 9.6% at level B, and 15.8% at level C per cardiac cycle (P<.001 for all levels). Significant diameter changes during cardiac cycles were also observed at all three levels. Pre-EVAR mean diameter changed up to 8.9% (P<.001) compared with post-evar aortic changes of up to 11.5% (P<.001). EVAR had no effect on change in aortic area and diameter. Dynamic MRA also demonstrated that pulsatile aortic distension was not equal in all axes, but rather occurred as an asymmetrical expansion and contraction. Conclusion: In patients with (atherosclerotic) aneurysm disease, the aortic dimensions at the level of and proximal to the aneurysm neck change during the cardiac cycle. This phenomenon is preserved after EVAR. Therefore, maximum diameter using dynamic MRA may not be similar to the maximum diameter with static CTA in all patients, and a standard regimen of 10% to 15% oversizing of an endograft based on static CTA images may be inadequate for some patients. Further studies using dynamic MRA to evaluate effects of different endografts are anticipated, with possible consequences for endograft designs. 96

3 Dynamic MRA: Conformational changes of the aneurysm neck Introduction Endovascular aneurysm repair (EVAR) has emerged from its infancy in 1991 to become the preferred treatment modality for appropriately selected patients with abdominal aortic aneurysms (AAA). 1 Properly selected patients with AAAs treated with EVAR can expect improved outcomes, a shorter hospital stay, and less surgical morbidity in the early postoperative period compared with conventional open surgery. 2,3 Patient selection has been based primarily on anatomic considerations, with most attention directed towards proper sizing of the endograft within the infrarenal aneurysm neck. 4,5 Most preoperative imaging protocols use computerized tomography angiography (CTA) with threedimensional (3D) reconstructions for sizing and planning. 6,7 However, several centers have reported successful use of magnetic resonance angiography (MRA) in preoperative EVAR planning with similar results to that of CTA. 8,9 Regardless of modality, the resulting images are static images, irrespective of the obvious fact that the human aorta exists in a dynamic environment. Aortic compliance and cardiac pulsatility naturally result in conformational changes during the cardiac cycle. 10,11 With the current high-speed multislice CT scans, the time it takes to scan the neck of an aneurysm only takes a fraction of the cardiac cycle. The static images acquired may represent the aortic neck during diastole (minimum diameter) or systole (maximum diameter), or somewhere in between. Subsequent sizing decisions are then made from these image measurements. The presence of significant pulsatile aortic neck variation may have serious implications for EVAR design, durability, and desirability. A potential risk of improper endograft sizing, with subsequent graft migration, intermittent type I endoleak, and poor patient outcome, exists. New, dynamic imaging tools are emerging to assess preoperative and postoperative aortic dynamics. With these tools, the effect of placing endografts of varying columnar strength into a relatively mobile, pulsatile aortic environment can be evaluated in an effort to improve stent-graft durability and results. The purpose of this study was to use highresolution electrocardiograph (ECG)-gated cine MRA to characterize aortic pulsatility during the cardiac cycle at important anatomic aortic landmarks in preoperative and postoperative AAA patients undergoing EVAR. Patients and Methods Patients Eleven consecutive patients were selected for EVAR and recruited into the study. All were men, with a median age of 74 years (range, 63 to 78 years). The study design and protocol were approved by the institutional medical ethics committee. Informed consent was obtained from all participants. 97

4 Imaging All scans were performed on a 1.5-T MR scanner (Gyroscan Intera, Philips Medical Systems, Best, the Netherlands). After initial multistack abdominal survey scans, a balanced fast field echo survey scan was performed to localize the renal arteries and aortic aneurysm. Transverse high-resolution scans with retrospective ECG gating were then obtained perpendicular to the long axis of the aorta at three levels. Level A was 1 cm below the lowest renal artery, level B was between the renal arteries, and level C was 3 cm above the lowest renal artery (Fig 1). The acquired voxel size was 2.1 x 0.78 x 6.0 mm 3, with a field of view of 400 x 320 mm 2 and using a scan percentage of 267% in the anteroposterior direction. The reconstructed voxel size was 0.78 x 0.78 x 6.0 mm 3, obtained with a reconstructed matrix of 512 x 512 pixels. Scan duration for obtaining a data set of 16 heart phases within each cardiac cycle was approximately 4 minutes at each level. Figure1. Levels at which transverse dynamic MRA scans were obtained. Level A, 1 cm below the lowest renal artery; level B, between the renal arteries; levelc, 3 cm above the lowest renal artery. Preoperative imaging MR scans were acquired in all patients the day before surgery. Sizing measurements were performed perpendicular to the central lumen line using preoperative static CTA with 20% oversizing according to institutional protocol. EVAR was performed by one surgeon using the Talent (Medtronic, Santa Rosa, CA, USA) or Excluder (W. L. Gore & Assoc, Flagstaff, AZ, USA) stent-graft systems. Postoperative MR scans were obtained 1 day after surgery (Fig 2). Preoperative and postoperative scans were technically successful, and cine loop images were obtained for all levels in each study patient. Image quality was considered to be excellent for all 11 patients. 98

5 Dynamic MRA: Conformational changes of the aneurysm neck Figure 2. Preoperative and postoperative representative magnetic resonance angiography images at level A (A and B) and level C (C and D). The presence of the endograft does not affect image quality. The left side demonstrates preoperative images. The right side demonstrates postoperative images. Analysis Dynamix software (Image Sciences Institute, Utrecht, the Netherlands) was used to analyze the dynamic scans. This software was developed to perform automated segmentation and measure changes in area and diameter at predetermined aortic levels (Fig 3). Each segmentation was independently reviewed manually by two blinded observers, and minor adjustments in the segmentation for small irregularities were required in approximately 35% of the images. Areas and minimum and maximum diameter along 256 axes, equallyspaced and through the center of mass of the aortic lumen, were also calculated during cardiac cycles. Statistical analysis of changes in area and diameters were performed using a Student s t- test for paired data. Significance was assumed at P<.05. Data on area and diameter were expressed as mean ± standard deviation. Analysis of measurement method comparison data according to Bland and Altman were performed to analyze repeatability and to compare measurements by two observers as well as comparing dynamic MRA and CTA data

6 Figure 3. Representative preoperative (A) and postoperative (B) images with the segmentation overlay shown. Results Aortic area Preoperative aortic area changed significantly during each cardiac pulsation at each of the three anatomic levels: above, at the level of, and below the renal arteries. At level A (within the aneurysm neck), the aortic area changed from 476 ± 144 mm 2 to 512 ± 143 mm 2 per cardiac cycle (P<.001); at level B (between the renal arteries), it changed from 471 ± 70mm 2 to 514 ± 71mm 2 (P<.001); and at level C (3 cm above the renal arteries), it changed from 475 ± 86 mm 2 to 535 ± 88 mm 2 (P<.001) (Fig 4). This corresponded to a pre- EVAR aortic area increase of up to 13.3% per cardiac cycle. The intraobserver repeatability coefficients were 22 mm 2 for observer 1 and 16 mm 2 for observer 2. The interobserver repeatability coefficient was 29 mm 2. Intraobserver and interobserver variability showed no significant differences within or between the observers. The postoperative aortic area also changed significantly during each cardiac cycle: from 472 ± 185 mm 2 to 516 ± 194 mm 2 at level A, from 439 ± 66 mm 2 to 480 ± 66 mm 2 at level B, and from a low area of 437 ± 87 mm 2 to a maximum area of 503 ± 90 mm 2 at level C (P<.001 for all levels) (Fig 4). This corresponded to a post-evar aortic area increase of up to 15.8% per cardiac cycle, which was not statistically different from pre-evar aortic area changes. Endograft placement did not significantly alter mean area change at any of the levels (Fig 5). The intraobserver repeatability coefficients were 39 mm 2 for observer 1 and 36 mm 2 for observer 2. The interobserver repeatability coefficient was 55mm 2. Again, intraobserver and interobserver variability showed no significant differences within or between the observers. 100

7 Dynamic MRA: Conformational changes of the aneurysm neck Minimum Maximum * * * * * * Area (mm 2 ) pre A post pre post pre post B C Anatomic Level Figure 4. Minimal and maximal aortic area during cardiac cycles measured preoperatively and postoperatively at three anatomic levels (A, B, and C). Data are means ± SD. *P < Pre EVAR Post EVAR 20 Change in Area (%) A B Anatomic Level C Figure 5. Preoperative and postoperative mean changes in aortic area during cardiac cycles at three anatomic levels (A, B, and C). Data are means ± SD. 101

8 Aortic diameter Aortic diameters changed significantly during the cardiac cycle at all measured levels. Preoperatively, mean aortic diameter changed from 23.4 ± 3.6 mm to 25.0 ± 3.5 mm at level A (P<.001), from 23.2 ± 1.8 mm to 24.8 ± 1.8 mm at level B (P<.001), and from 23.6 ± 2.2mm to 25.6 ± 2.2 mm at level C (P<.001) (Fig 6). At level A, mean diameter changes over 256 axes ranged between 0.6 ± 0.3 mm and 3.6 ± 0.6 mm. Similar diameter changes were observed at level B (0.3 ± 0.4 mm to 3.6 ± 1.0 mm) and at level C (0.7 ± 0.2 mm to 4.2 ± 0.9 mm). The intraobserver repeatability coefficients were 0.6 mm for observer 1 and 0.8 mm for observer 2. The interobserver repeatability coefficient was 1.0 mm. Intraobserver and interobserver variability showed no significant differences within or between the observers. 35 Minimum Maximum * * * * * * Diameter (mm) pre post pre post pre post A B C Anatomic Level Figure 6. Preoperative and postoperative minimum and maximum aortic diameters in 256 axes obtained during cardiac cycles at three anatomic levels (A, B, and C). Data are means ± SD. *P < Postoperatively, mean aortic diameter also changed significantly during each cardiac cycle. At level A, it changed from 22.9 ± 4.4 mm to 25.1 ± 4.4 mm; at level B, it changed from 22.4 ± 1.8 mm to 24.3 ± 1.8 mm; and at level C, it changed from 22.8 ± 2.3 mm to 25.2 ± 2.0 mm (P<.001 for all levels) (Fig 6). Within the 256 different axes, diameters changed between a minimum of 0.8 ± 0.5 mm and a maximum of 4.5 ± 1.0 mm at level A, between 0.4 ± 0.5 mm and 4.2 ± 0.8 mm at level B, and between 1.0 ± 0.4 mm and 4.6 ± 0.8 mm at level C. This change corresponded with an increase in aortic diameter of up to a 22%. Post-EVAR diameter changes were not different from pre-evar changes. Stentgraft placement did not significantly alter mean diameter change at any of the levels (Fig 7). The intraobserver repeatability coefficients were 1.7 mm for observer 1 and 1.0 mm for 102

9 Dynamic MRA: Conformational changes of the aneurysm neck observer 2. The interobserver repeatability coefficient was 1.8 mm. Again, intraobserver and interobserver variability showed no significant differences within or between the observers Pre EVAR Post EVAR 12 Change in Diameter (%) A B Anatomic Level C Figure 7. Preoperative and postoperative mean changes of aortic diameters in 256 axes during cardiac cycles at three anatomical levels (A, B, and C). Data are means ± SD. Aortic diameter using dynamic MRA vs static CTA (before EVAR) The mean maximum diameter for the 11 patients before EVAR showed no significant differences comparing dynamic MRA and static CTA (mean difference, 0.9 mm ± 2.6 mm; P =.29). In one patient, however, a significantly higher maximum diameter (6.6 mm) was seen with dynamic MRA, even exceeding a 20% oversizing of the maximum aortic diameter based on static CTA. During followup, a proximal type I endoleak developed in this patient. A type I endoleak was also seen in one other patient. The maximum diameter in this patient showed no significant difference when measurements with dynamic MRA and static CTA were compared. Discussion To our knowledge, this is the first publication of the use of cine MRA to study the dynamic effects on the aneurysm neck before and after endograft placement for the treatment of AAAs. High spatial resolution allows precise imaging and subsequent measurements. By using retrospective ECG gating, time-resolved images at clinically relevant aortic 103

10 levels can be acquired, resulting in cine MRA loops with 16 images per cardiac cycle. We therefore have achieved excellent temporal and spatial resolution. This new powerful tool allows one to view dynamic changes in the aorta with every beat of the heart. The potential implications of this tool could be enormous to vascular surgeons and engineers contemplating endograft design, durability, and potential pitfalls. This imaging method has some limitations. In addition to the standard contraindications to MRA, only patients with a specific type of endograft can be evaluated with dynamic MRA. All of our patients received the Talent or Excluder devices, which are compatible with MRA. Our institution has previously evaluated the MRI characteristics of various endografts. 13 Of those studied, the AneuRx (Medtronic), Talent, Excluder, and Quantum LP (Cordis, Warren, NJ) stent-grafts appear to be amenable to MRI evaluation, whereas the ferromagnetic properties of the Zenith (Cook, Bloomington, Ind) and Lifepath (Edwards Lifesciences, Irvine, Calif) devices resulted in large susceptible artifacts that precluded adequate image quality. 13 Therefore, patients with Zenith and Lifepath endografts are not currently candidates for this new imaging tool. All measurements were performed at predetermined, clinically relevant anatomic levels. We chose 3 cm above the lowest renal artery, at the level of the renal arteries, and 1 cm below the lowest renal artery. These levels corresponded to native aorta, aorta with suprarenal fixation, and aneurysm neck with endograft fabric, respectively. However, aortic wall motion obviously occurs at all levels and in all directions. This study is limited in evaluating only the three abovementioned levels in two dimensions. Three-dimensional movements might make the analyzed cross-sectional area in our study move slightly out of plane during the cardiac cycle. Theoretically, this could have some influences on our results. Three-dimensional volumetric analysis could give further interesting results, but this technique, which measures movement in all possible planes within a reasonable scan duration, is not yet available in our hospital. Aortic dynamics may play a role in the design of future endografts. Pulsatile forces with each heartbeat over a patient s lifetime can result in literally millions of repeated stress events placed upon an implanted endograft. Clinicians have witnessed stent fractures, fabric erosions, suture breakage, and endograft erosions through native arteries Although this report does not evaluate endograft durability, it does characterize and quantify the dynamic environment in which endografts are placed. As the frontiers of endografts design are pressed forward with the development of fenestrated and branched endografts, the devices and the dynamics become increasingly complex. 3,19,20 Cine MRA is a new tool that could be used to evaluate and potentially improve future designs. Computer analysis enabled us to determine the greatest change in diameter at each predetermined aortic level, simultaneously along 256 axes, equally-spaced and through the center of mass of the aortic lumen during the cardiac cycle. At the same level and during the same cycle, aortic diameter can change by as much as 4.2 mm or as little 0.3 mm, depending on which axis is being viewed. This may imply an asymmetrical expansion of the aorta during systole. 104

11 Dynamic MRA: Conformational changes of the aneurysm neck Data analysis showed a significant change in aortic diameter and in aortic area, both preoperatively and postoperatively during the cardiac cycle at each level. In the studies by Vos et al, 21 pulsatile aneurysm wall motion was negligible before and after EVAR. The measurements in their studies were performed at the level of the aneurysm sac, however, not at the level of the undilated aorta proximal to the aneurysm. 21,22 Furthermore, in our study intraobserver and interobserver repeatability coefficients were relatively low. This could be the result of the high quality of the images and the use of Dynamix software that offered automated segmentation. Because the preoperative and postoperative changes in aortic diameter and area exceeded the above-mentioned repeatability coefficients, these changes can be considered significant and are therefore clinically relevant. No significant differences were seen between the preoperative change in aortic area or diameter (pulsatility) and the changes after EVAR. During the postoperative measurements, however, mean arterial blood pressure was significantly lower, which could have influenced these results. Although stent grafts were oversized by 20%, endografts placement did not significantly alter the change in aortic area or in maximum aortic diameter at any of the levels, and two-dimensional pulsatile aortic wall motion was preserved. This contrasts with previous studies that showed a reduction in aortic diameter and pulsatile wall motion after EVAR. 10,23 However, in these studies evaluations were performed by ultrasound imaging, 10 or on a small group of patients being evaluated before EVAR compared with a large group of patients with endoleaks after EVAR. 23 Again, measurements (before and after endograft placement) in both studies were performed at the level of the aneurysm sac 10,23. In fact, to our knowledge, our study is the first to use dynamic MRA in quantifying aneurysm pulsatility at three levels encompassing the aneurysm neck, as mentioned earlier. Previous studies on aneurysm neck wall movement were either not dynamic, 24 were invasive, contained only small groups of patients, or were in vitro measurements. 25 The preoperative maximum aneurysm neck diameter comparing dynamic MRA and static CTA showed a significantly higher diameter in one patient that exceeded an oversizing of 20% based on static CTA. A proximal type I endoleak developed in this patient during follow-up. Endograft oversizing of 20% based on static CTA may not have been adequate in this patient. 105

12 Conclusion This study introduces the feasibility of cine MRA imaging on dynamic aortic wall motion before and after EVAR at the level of the undilated aorta proximal to the abdominal aneurysm. Understanding the dynamic properties in this area could have important affects on stent-graft fixation and design. Patients with AAAs selected for EVAR demonstrate changes in aortic diameter with each cardiac cycle. The native aorta exhibits significant pulsatility, and this phenomenon is preserved after endograft implantation. Morphologic changes are very complex, but this study gives early insight into changing aortic morphology per cardiac cycle. Future studies using dynamic MRA to determine rupture risk, effects of different endografts, volumetric analysis, and even consequences for endograft efficacy and durability are anticipated. 106

13 Dynamic MRA: Conformational changes of the aneurysm neck References 1 Parodi JC, Palmaz JC, Barone HD. Transfemoral intraluminal graft implantation for abdominal aortic aneurysms. Ann Vasc Surg 1991;5: Prinssen M, Verhoeven EL, Buth J, Cuypers PW, van Sambeek MR, Balm R, Buskens E, Grobbee DE, Blankensteijn JD; DREAM Trial Group. A randomised trial comparing conventional and endovascular repair of abdominal aortic aneurysms. N Engl J Med 2004;351: Greenhalgh RM, Brown LC, Kwong GP, Powell JT, Thompson SG; EVAR trial participants. Comparison of endovascular aneurysm repair with open repair in patients with abdominal aortic aneurysm (EVAR trial 1), 30-day operative mortality results: randomised controlled trial. Lancet 2004;364: Green RM. Patient selection for endovascular abdominal aortic aneurysm repair. J Am Coll Surg 2002;194: Schumacher H, Eckstein HH, Kallinowski F, Allenberg JR. Morphometry and classification in abdominal aortic aneurysms: patient selection for endovascular and open surgery. J Endovasc Surg 1997;4: Sprouse LR 2nd, Meier GH 3rd, Parent FN, DeMasi RJ, Stokes GK, LeSar CJ, Marcinczyk MJ, Mendoza B. Is three-dimensional computed tomography reconstruction justified before endovascular aortic aneurysm repair? J Vasc Surg 2004;40: Broeders IA, Blankensteijn JD, Olree M, Mali W, Eikelboom BC. Preoperative sizing of grafts for transfemoral endovascular aneurysm management: a prospective comparative study of spiral CT angiography, arteriography, and conventional CT imaging. J Endovasc Surg 1997;4: Neschis DG, Velazquez OC, Baum RA, Roberts D, Carpenter JP, Golden MA, Mitchell ME, Barker CF, Pyeron A, Fairman RM. The role of magnetic resonance angiography for endoprosthetic design. J Vasc Surg 2001;33: Wyers MC, Fillinger MF, Schermerhorn ML, Powell RJ, Rzucidlo EM, Walsh DB, Zwolak RM, Cronenwett JL. Endovascular repair of abdominal aortic aneurysm without preoperative arteriography. J Vasc Surg 2003;38: Malina M, Lanne T, Ivancev K, Lindblad B, Brunkwall J. Reduced pulsatile wall motion of abdominal aortic aneurysms after endovascular repair. J Vasc Surg 1998;27: Duvernoy O, Coulden R, Ytterberg D. Aortic motion: a potential pitfall in CT imaging of dissection in the ascending aorta. J Comput Assist Tomogr 1995;19: Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1: van der Laan MJ, Bartels LW, Bakker CJ, Viergever MA, Blankensteijn JD. Suitability of 7 aortic stent-graft models for MRI-based surveillance. J Endovasc Ther 2004;11: Krajcer Z, Howell M, Dougherty K. Unusual case of AneuRx stentgraft failure two years after AAA exclusion. J Endovasc Ther 2001;8: Teutelink A, van der Laan MJ, Milner R, Blankensteijn JD. Fabric tears as a new cause of type III endoleak with Ancure endograft. J Vasc Surg 2003;38: Boyle JR, Thompson MM, Clode-Baker EG, Green J, Bolia A, Fishwick G, Bell PR. Torsion and kinking of unsupported aortic endografts: treatment by endovascular intervention. J Endovasc Surg 1998;5:

14 17 Bockler D, von Tengg-Kobligk H, Schumacher H, Ockert S, Schwarzbach M, Allenberg JR. Late surgical conversion after thoracic endografts failure due to fracture of the longitudinal support wire. J Endovasc Ther 2005;12: Malina M, Brunkwall J, Ivancev K, Lindblad B, Malina J, Nyman U, Risberg B. Late aortic arch perforation by graft-anchoring stent: complication of endovascular thoracic aneurysm exclusion. J Endovasc Surg 1998;5: Verhoeven EL, Prins TR, Tielliu IF, van den Dungen JJ, Zeebregts CJ, Hulsebos RG, van Andringa de Kempenaer MG, Oudkerk M, van Schilfgaarde R. Treatment of short-necked infrarenal aortic aneurysms with fenestrated stent-grafts: short-term results. Eur J Vasc Endovasc Surg 2004;27: Greenberg RK, Haulon S, Lyden SP, Srivastava SD, Turc A, Eagleton MJ, Sarac TP, Ouriel K. Endovascular management of juxtarenal aneurysms with fenestrated endovascular grafting. J Vasc Surg 2004;39: Vos AW, Wisselink W, Marcus JT, Manoliu RA, Rauwerda JA. Aortic aneurysm pulsatile wall motion imaged by cine MRI: a tool to evaluate efficacy of endovascular aneurysm repair? Eur J Vasc Endovasc Surg 2002;23: Vos AW, Wisselink W, Marcus JT, Vahl AC, Manoliu RA, Rauwerda JA. Cine MRI assessment of aortic aneurysm dynamics before and after endovascular repair. J Endovasc Ther 2003;10: Faries PL, Agarwal G, Lookstein R, Bernheim JW, Cayne NS, Cadot H, Goldman J, Kent KC, Hollier LH, Marin ML. Use of cine magnetic resonance angiography in quantifying aneurysm pulsatility associated with endoleak. J Vasc Surg 2003;38: Raghavan ML, Vorp DA, Federle MP, Makaroun MS, Webster MW. Wall stress distribution on three-dimensionally reconstructed models of human abdominal aortic aneurysm. J Vasc Surg 2000;31: Flora HS, Woodhouse N, Robson S, Adiseshiah M. Micromovements at the aortic aneurysm neck measured during open surgery with closerange photogrammetry: implications for aortic endografts. J Endovasc Ther 2001;8: